R53 Routing Policies Summary

Policies with 1-2 records per host name (simple policies)

R53 Routing Policy	Architecture	Records returned	Use cases
Simple	- One record per host name, multiple values per record

No health checks | All record values returned to client in random order, client chooses one | - Route requests towards one single service (e.g. web server)
Very easy to manage | | Failover | - Two records with same host name: primary and secondary resources
Health check for primary resource | - Primary resource healthy → primary record
Primary resource unhealthy → secondary record | Active-passive failover. E.g. failover for an out-of-band failure or maintenance page |

R53 Routing Policy	Architecture	Records returned	Use cases
Multivalue answer	Multiple records with same host name, all valued equally	Up to 8 healthy records (chosen randomly)	Active-active approach to DNS, improves availability
Weighted	Records have weights	Healthy records returned on a frequency based on their record weight vs total weight	- Test new versions of SW

Very simple load balancing | | Latency | Records have different AWS regions | Healthy record with lowest latency to client | Optimize performance and UX of global applications | | Geolocation | Records have different location tags | Healthy record that is most specific for the user location (subdivision → country → continent → default/no answer) | - Content restriction
Language specific content
Load balancing across regional endpoints based on customer location | | Geoproximity | Records have geographic location and (optionally) bias | Healthy record that is closest geographically to the client, taking bias into account | Flexible routing policy that can account for proximity and artificial bias |